1

WORK MEASUREMENT FOR ESTIMATING FOOD PREPARATIONTIME OF A BIOREGENERATIVE DIET

Ammar Olabi1*, Jean Hunter2, Peter Jackson3, Michele Segal4, Rupert Spies5,

Carolyn Wang6, Christina Lau6, Christopher Ong6, Conor Alexander6, Evan

Raskob6, Jennifer Plichta6, Ohad Zeira6, Randy Rivera6, Susan Wang6, Bill

Pottle6, Calvin Leung6, Carrie Vicens6, Christine Tao6, Craig Beers6, Grace Fung6,

Jacob Levine6, Jaeshin Yoo6, Joanna Jackson6, Kelly Saikkonen6, Matthew

Zimmerman6, Megan Cunningham6, Michele Crum6, Naquan Ishman6, Norman

Voo6, Raul Cadena6, Robert Relinger6, Saori Wada6

1Food Science Department, Stocking Hall, Cornell University, Ithaca NY 14853

2Biological and Environmental Engineering Department, Riley-Robb Hall,Cornell University, Ithaca NY 14853

3School of Operations Research and Industrial Engineering, Rhodes Hall, CornellUniversity, Ithaca NY 14853

4Design and Environmental Analysis Department, Martha Van Rensselaer Hall,Cornell University, Ithaca NY 14853

5School of Hotel Administration, Statler Hall, Cornell University, Ithaca NY14853

6Students in the ABEN/ORIE/HA 499 tutorial class during the Fall 1998-Spring2000 period, Cornell University, Ithaca NY 14853

*Corresponding Author: Ammar Olabi 1 Grand Avenue California Polytechnic State University San Luis Obispo CA 93407 Phone: (805) 756-2997(work) (805) 596-0768 (home) Fax: (805) 756-1146 E-mail: AOLABI@CALPOLY.EDU

2

ABSTRACT

During space missions, such as the prospective Mars mission, crew labor time is a

strictly limited resource. The diet for such a mission (based on crops grown in a

bioregenerative life support system) will require astronauts to prepare their meals

essentially from raw ingredients. Time spent on food processing and preparation

is time lost for other purposes. Recipe design and diet planning for a space

mission should therefore incorporate the time required to prepare the recipes as a

critical factor. In this study, videotape analysis of an experienced chef was used to

develop a database of recipe preparation time. The measurements were highly

consistent among different measurement teams. Data analysis revealed a wide

variation between the active times of different recipes, underscoring the need for

optimization of diet planning. Potential uses of the database developed in this

study are discussed and illustrated in this work.

KEYWORDS: diet, food preparation, space, time studies, work measurement

3

CONTENT SENTENCE: Work measurement videotaping techniques are used

to collect labor time data on food preparation activities of a bioregenerative life-

support space diet.

4

INTRODUCTION

During space missions, such as the prospective Mars mission, the availability of

crew labor time is a strictly limited resource (11). Any tool used to increase task

efficiency or to make the schedule of astronauts more manageable will benefit

both astronauts and mission planners. On current Shuttle missions, the crew time

devoted to preparation and cleanup of prepackaged food ranges from 45 to 90

minutes/day total (9). In a bioregenerative life support system (BLSS), astronauts

will have to prepare their meals using the crops grown in a hydroponic system.

Such meal preparation from basic ingredients will consume far more of the crew's

time than currently experienced. Accurate estimates of the crew time required for

food preparation are needed to support NASA’s decisions on food system choices

for long duration missions. The economic component of those decisions is likely

to use equivalent system mass (ESM) analysis.

Equivalent system mass (ESM) is a measure that converts physical units of non-

mass resources such as power, pressurized volume and crew labor into equivalent

payload mass by multiplying the units with appropriate mass penalties. ESM

accounting of plans for space missions allows for the comparison among different

life support system options using a single unit of measure: the ESM of the option

(6, 10). The ESM cost of labor [kg/work hour] will depend on the ESM of the life

support system, the number of hours in a standard workweek and other life

5

support demands on crew time.

The first step in the development of models for predicting work-times during a

space mission is timing of the tasks on the ground. For example, in the

development of crew schedules for Space Shuttle experiments, ground

simulations were used to predict the labor requirement of each step in an

experiment (Johnson, C. personal communication). Videotape analysis techniques

are well suited to this task and have been used by NASA to document space usage

by the crew during the LMLSTP ІІІ closed chamber test (21).

This paper presents a method for estimating food preparation time using

videotape analysis and reviews the results of applying this method to the time

study of a small group of recipes suitable for a bioregenerative diet.

We suggest that NASA apply work measurement techniques to ground

simulations of food preparation and other life support activities as a prelude to

planning tasks during future space and planetary missions. Together with

ergonomics, this information can help in designing the food preparation area for

future missions. We considered food preparation for only planetary surface

missions since food processing and preparation is very difficult in microgravity.

6

Model for Current Study

This project described in this paper was part of a larger study aimed at optimizing

the crew diet for a bioregenerative planetary surface mission (12). Food

preparation labor must be estimated in order to include it as a cost in the system.

For this labor study, each recipe is analyzed into its component tasks. Food

processing labor and food preparation labor are calculated and separately assigned

to the processing costs of ingredients and the preparation cost of recipes,

respectively. The ESM cost estimation procedures used are outlined in (6, 10).

History of Work Measurement Techniques

Measurement and analysis of work has been used since the industrial revolution to

improve worker productivity and comfort. Robert Owen, the first to analyze work

as a process, identified the need for rest and fatigue allowances for workers, and

showed that output was not directly proportional to time worked (26).

Contemporary work measurement was founded at the turn of the century by

Frederick Taylor. He created a technique for standardizing performance times of

repetitive operations by specifying the method of operation and by timing the

component tasks with a stopwatch (13).

In the 1920’s, sampling theory made possible more effective data collection and

the use of larger sample sizes. In the 1930’s, Frank and Lillian Gilbreth

7

established motion study, used to simplify and speed up assembly line tasks. The

Gilbreths developed micromotion cinematography to take precise and consistent

measurements of tasks. They subdivided repetitive tasks into micromotions, such

as ‘reach’, ‘grasp’, and ‘position’, allowing for detailed analysis of manual work.

Their data, among others, were used to create “predetermined motion-time

system” models which predict the time requirement for a given task from the

times required for its micromotions. Work analysis remains a key tool of

manufacturing engineering (17) and has been applied to issues of cost

management as well as assembly line operation (26).

Types of Work Measurement Techniques

Four major techniques are used for work measurement: Time Study, Standard

Data, Predetermined Time Systems, and Work Sampling (2). Usually the

objective of work measurement is the development of a standard time, defined as

the time required by an average worker, working at a normal pace, to complete a

specified task using a prescribed method. The tasks of interest here are the

preparation of various recipes by a cook with average kitchen skills, who is

familiar with the recipe and with the kitchen and pantry layout.

In Time Studies, data on task duration is collected using either a stopwatch or a

video camera. The recording speed of the camera can be adjusted depending on

the amount of detail needed. The task under study must have a well-defined

8

beginning and end. It should be measured only under typical situations, and for

privacy and wage negotiation reasons, all the parties involved in the study should

be informed in advance (22). The task should have a relatively short duration but

have sufficient duration to be accurately timed.

Predetermined Time Systems (PTS) methods rely on a systematic procedure

which subdivides any complex manual task into motions, body movements, or

other elements of human performance, and assigns to each element the

appropriate time value from a table of time/motion data.

The third main method of work measurement is Standard Data. The inputs for a

“standard data” analysis come from time studies and/or predetermined time

system estimates (22). Standard data systems organize standard times from a

number of related tasks into a database from which the standard times for related

jobs may be predicted (1); in essence, it is a means to adjust or extrapolate PTS or

time studies results to new tasks.

Work Sampling is a statistical method where an analyst records, at random

intervals, the type of task the worker is doing. The number of observations

recording a particular task reveals the fraction of worker time devoted to that task.

Statistical formulas are used to determine the correct number of work samples to

record based on rough estimates of the duration and frequency of each task (1).

This method is usually applied to investigate the division of worker time among

9

the various activities that comprise a job or a work situation (22), and is best

suited to jobs of long duration and low repetition.

The choice of a work measurement technique depends on several factors: the

nature of the job being analyzed (its complexity and use of repetitive tasks), the

number of people involved, time and budget constraints, and the objectives of the

study (7).

Work Measurement in Food Production

In the food industry, work measurement is applied to improve productivity in

industrial processing or in food service level contexts such as hospitals or chain

restaurants (8,18). For example, Time and motion studies were used by Burger

King management to cut transaction time to 30 seconds at the drive-through

service window (15).

Hospital dietary units have used work measurement to identify opportunities for

productivity improvement. Donaldson (5) used work sampling to measure the

productivity of food production personnel and the labor time per meal. Blanken

(3) first measured tasks, then improved productivity by reallocating the tasks

among personnel according to relevant time or employee constraints.

Motion and time studies, as well as sampling techniques, have been successfully

10

employed to isolate costly functions and to improve the efficiency of kitchen

workers (19, 27, 4, 24). Work sampling has also been used to measure

supervisory and managerial work in a kitchen setting (20, 23). A study by

Montag (16) aimed at assessing the use of predetermined motion time systems in

order to predict the production times for specific foods. The author concluded that

while consistent estimates could be made, the time needed to conduct the analysis

was too long and beyond economical benefit.

MATERIALS AND METHODS

From the available methods of work analysis, the time study method was chosen

because the work performed included the preparation of many different recipes

with very little repetition. Although a predetermined time system would have

worked, it would have required too fine a level of detail and would not have been

cost effective. Work sampling is often applied to activities with relatively long

cycle and a low degree of repetition such as some of our cooking tasks, but it

requires a large number of observations, impractical in our case since each recipe

was filmed only once. Although each recipe was filmed only once, the individual

tasks were replicated many times across the different recipes.

11

Data Collection

Video data were collected using two camcorders (Sony Video 8XR model

CCDTRV15) positioned near the ceiling on opposite sides of the food preparation

room and focused on the five major workstations: central island, sink, stove, food

processor and oven. Camera positioning allowed for an overlapping view of the

food preparation area. The camcorders were equipped with time stamps and dates,

and were synchronized to the nearest second. The microphones on the two

camcorders had good sound quality, necessary for the chef who was required to

verbally identify any steps, tasks or ingredients that might be difficult to identify

in the video analysis. The camcorders fed individually into two VCRs (Panasonic

model 8405S) which were linked to a Zenith monitor (model #A13PO2D) which

allowed the chef to check for proper camera function before starting work.

Resulting tapes were replayed using a VCR with a jog dial (Sony model SLV-

77Hi-Fi) which enabled the viewer-analyst to advance the action frame-by-frame

to accurately record task start and end times. The preparation sessions for a total

of 40 recipes were videotaped in this manner.

In all of the videotaped recipes, the person videotaped was the chef (a research

support specialist) on the project. The chef had long experience in food

preparation and was preparing each of the recipes for the second time when the

videotaping occurred. This corresponds somewhat to the astronaut situation

because the astronauts will be expected to have some culinary training and will

12

become familiar with the recipes as they prepare them. On the other hand, we

recognize that skill level can have a large impact on food preparation times. For

example, a preliminary study (R. Spies and R. Relinger, 2000), not reported here,

of three subjects (inexperienced, trained, and professional) revealed differences in

task times as high as 1000% for difficult tasks such as slicing mushrooms. For

consistency, our video data are based on a single skill level. The only criterion

adopted for the selection of recipes for filming was to cover a wide range of

different food categories such as breakfast foods, main dishes, soups, appetizers,

etc.

Videotape Analysis

Video analysis in this project was a 3-step process. For each recipe, a group of

task analysts initially created a list of expected tasks based on the recipe, and then

edited this list during a first viewing of the videotape, adding additional tasks and

noting videotape segments that would be classified as time-off tasks, failed

actions, or mistakes (such as dropping a piece of food). The finished task lists

were passed along to a second group of two viewers who recorded the time

elapsed for each task to the nearest second, using the time stamp on the video, and

entered the data into a database.

A key concern in developing the methodology was to ensure consistent

13

classification and measurement of videotape events by a diverse collection of

volunteer videotape analysts. Two types of inconsistencies were anticipated: first,

that different analysts would classify similar activities in different ways, and

second, that task durations would show significant between-analyst variation. To

guard against these kinds of inconsistencies, as well as to distribute the workload,

we separated the classification analysis from the measurement analysis. Each

recipe was coded (analyzed into tasks) by exactly one task definition team and

measured, during the preliminary analyses, by at least two task measurement

teams. To guard against the first inconsistency, the task definition teams met

frequently as a group to create a standard list of tasks and to resolve classification

issues. To guard against the second inconsistency, the task definition team

produced a set of tasks (Appendix A) and identification guidelines for the

measurement teams.

Task List Development

The goals of task list development were to ensure coverage of all activities

occurring in the 40 videotaped recipes, to avoid overlap between task definitions

created, and to make the list short and simple. The tasks were defined to be easily

understood by students on the task measurement teams. A list of 55 cooking tasks

(including dead time) was first developed from the recipe sheets of 40-videotaped

recipes (Appendix A). Representative tasks are shown in Table 1. The recipes

14

used in the study typically had a batch size of 6 to 12 servings. Identification of

the start and end times for each task were defined in terms of the tools used. Thus,

the task began whenever the equipment or ingredient was grasped by the chef and

ended when it was released. This approach greatly reduced the problem of

ambiguity in analyzing task durations and contributed to the consistency found

among measurements by different analysts. Since the duration of transits between

workstations varies with the size and design of the kitchen, the time spent on

transits was not included in the total time of the particular task, but was called out

separately as a transit.

Four general task categories were defined. The first and most important was that

of active tasks such as peeling, washing, measuring, cutting, assembly, and

various types of cooking tasks. The second category consisted of passive tasks

which did not require continuous involvement. Examples included tasks such as

boil, cook, bake, and preheat. Active tasks could be performed during execution

of a passive task. The third category was transits, that is, movement between

workstations.

INSERT TABLE 1 AROUND HERE

The fourth category was dead time, accounting for interruptions such as phone

calls or tasks for another recipe. Two additional tasks, the weighing of ingredients

and the recording of measurements (writing) in the lab notebook, were recorded

15

but not used in analysis because they would not be part of ordinary food

preparation. Typically in food preparation, most of the ingredients are measured

rather than weighed but weighing was done in this study for the sake of accuracy

and other nutritional analyses that were required.

Database

A database was created in Microsoft Access 97 to hold the timing data for the

different recipes. The database was designed to be user friendly, easily searchable

by data queries, and expandable, with options of adding new dishes, new task

actions and new task objects as necessary. Scrolling lists in the different tables

were designed to minimize typing during data entry. The database structure,

including variables used, the tables capturing these variables, and the relationships

among the tables, is diagrammed in Figure 1. Note that each table has one or more

links with the main table (Table Task and Times), into which the data were

entered. The main variables for each task were Task Action, Task Object, Start

Time, End Time, and comments (Table 2). Comments were added where

necessary by the task definition team to facilitate data input, or by the data timing

group in case of confusion or error.

INSERT FIGURE 1 AROUND HERE

16

INSERT TABLE 2 AROUND HERE

RESULTS AND DISCUSSION

We initially evaluated the methods and data of this study based on six food

preparation events, each for a different recipe. These were taped and analyzed by

the methods described earlier. Total recipe preparation time was divided among

active food preparation time, passive time, dead time, transit time, weighing

ingredients, and writing. Figure 2 displays the breakdown for each recipe. We

focused on the active and passive food preparation time components because the

other components offer little prediction value.

The first conclusion of the preliminary analysis was that between-analyst

(different timing groups) task time variability was not significant (p > 0.05) as

obtained from a two-way ANOVA. The two-way test was performed on five of

the six recipes. (Asian Pesto was not included because it was measured by only

one team.) The two factors considered in the ANOVA were the task number

("Task") and the team number ("Group"). While task factor was significant in all

five recipes (p-value < 0.05), the group factor was not significant (p-value > .14).

In particular, it was not significant for the one recipe measured by all four teams

(Melon Ginger: p-value > 0.89).

17

INSERT FIGURE 2 AROUND HERE

Since the group factor was not significant, videotape analysis was continued on

10 additional recipes without replicating task time measurements. Figure 3 shows

the breakdown of the 10 additional recipes into their time category components.

Preliminary analysis showed that the discarded time components (transits,

weighing and writing) averaged 17.6% of the total recipe preparation time (Soy

Wheat Crepes excluded). As expected, a large variation in active food preparation

time was observed among the different recipes. Seitan Stir Fry, for example, has

four times the active preparation time of Peanut Wheat Hot Cereal. These

differences are likely to influence menu selection because of the high cost of

labor.

INSERT FIGURE 3 AROUND HERE

Separate accounting of active and passive time allowed observation of differences

between recipes that would have been obscured by grouping active and passive

tasks together. For example, had the passive and active time been counted

together in one time category then recipes like Potato Onion Bread and Tofu

Cottage Cheese would have significantly different labor requirements. However,

the active time is almost identical (37.37 and 40.48 minutes for Potato Onion

Bread and Tofu Cottage Cheese, respectively). The length of passive time

18

intervals and their sequence among active tasks is useful information when

planning simultaneous preparation of several recipes, and may also provide

information about adequate recipe combinations.

Tasks were also divided into two scaling categories: quantity dependent (QD) and

quantity independent tasks (QI). QI tasks, such as picking up and putting down

utensils, setting up equipment such as the food processor, or opening and closing

containers do not vary in duration according to the batch size of the recipe. On the

other hand, the duration of a QD task depends on the amount of material or

ingredient used in the task and, hence, on the batch size. Examples of QD tasks

are hand-chopping vegetables or fruits, rolling out dough, or peeling. The average

durations across all recipes for tasks in the quantity independent and quantity

dependent categories are shown in Figures 4 and 5 respectively.

INSERT FIGURE 4 AROUND HERE

INSERT FIGURE 5 AROUND HERE

On average, the QI tasks are much shorter in duration than the QD tasks. Fourteen

of 20 QI tasks average around or under 10 seconds. As for the QD tasks, 29 of 34

are under 100 seconds.

Figure 6 shows a cumulative distribution function graph for three common

19

quantity independent tasks: drain, check and close. The range of task times

observed for these tasks is shown on the vertical axis. The distribution function

captures the variation in task times observed. For example, approximately 75% of

the tasks last longer than 30 seconds and are concentrated in a range between 30

and 35 seconds. The remaining 25% of the observations are distributed evenly

over the broad range from 1 to 30 sec.

INSERT FIGURE 6 AROUND HERE

The ESM cost prediction; the required time per recipe can be estimated by

summing the mean times for all active tasks. For QI tasks, this will simply be the

average observed duration. For QD tasks, it will be an average task rate multiplied

by the quantity of material to be processed. The total active time should be

adjusted for transits, inflated by an appropriate dead time factor and added to the

cleanup time (computed separately) to yield a total time requirement.

The value of the timing data collected extends beyond its use in measuring labor

costs. Pareto analysis (25) of the task times for a recipe can lead to improved

work methods to economize on labor. Figure 7 displays a Pareto analysis for the

Seitan Stir Fry dish.

INSERT FIGURE 7 AROUND HERE

Table 3 lists the top ten tasks from this dish, in order of duration. The longest task

20

turns out to be cutting pepper, a manual activity. This was due to the fact that the

pepper had to be cut by hand into a combination of uniform thin slices and dice

chunks. This task could be performed more quickly in a food processor or it could

be shortened by specifying all slices, or all large chunks, as the shape of the food

pieces. Thus, if a recipe is initially rejected from the menu optimization as too

labor-intensive, modifications in the recipe or in work methods may permit it to

be reconsidered.

INSERT TABLE 3 AROUND HERE

Another potential use of work measurement data is to optimize meal preparation

activities, in situations in which several recipes are prepared simultaneously at a

limited number of workstations, with the goal of having the entire meal ready at

the same time. Such an optimization needs the following inputs: the time required

by the various tasks in the recipes of interest, the required sequence of these tasks,

and the resources needed for these tasks. Estimates for all of these parameters are

easily derived from the video analysis. Figure 8 shows an example of a Gantt

chart (14) to optimize the food preparation activities for a meal of 5 recipes. The

recipes were among the most complicated of the recipe set and were chosen to

show the use of multiple equipment stations and the challenges of sequencing

active tasks with passive intervals such as baking and chilling. Gantt charts,

named after Henry Gantt, are visual displays of task durations and sequences and

21

can be used to analyze causes of delays. The arcs in Figure 8 indicate the task

sequences. The workstations in the graph include the stove, oven, refrigerator,

sink, freezer, food processor, blender, and counter. The active time used in the

preparation of these 5 recipes was 3.5 hours. This unrealistic figure is due to the

use of conservative task labor time estimates. Time study data were not available

for the complicated recipes chosen.

INSERT FIGURE 8 AROUND HERE

As Figure 9 shows, most of the time needed for the production of the 5 recipes

involved the use of the counter (center island), followed in order of usage by the

stove and then the food processor.

INSERT FIGURE 9 AROUND HERE

The results of this type of analysis could lead to improvements in design of the

food preparation area, automation of repetitive tasks or changing the capacity of

equipment. For example, if work is frequently delayed because of limited counter

space the counter space could be increased. Likewise, the manual tasks most in

need of mechanization can be identified.

22

CONCLUSIONS

Estimating food preparation labor time is an important factor in optimizing menu

selection for a bioregenerative diet. The food preparation tasks used in the recipes

of this work should be, in their majority, adaptable to a planetary colony scenario.

Recipe-specific labor requirements were measured by videotaping an experienced

chef preparing the recipes, and then classifying and timing the distinct tasks

required to prepare the recipe. This methodology produces consistent and accurate

task duration estimates. Statistical analysis of the task durations found no

significant effects due to data analyst group. As expected, preparation times

varied considerably across recipes, underscoring the importance of including

labor cost in a menu/diet optimization. Task analysis revealed opportunities to

improve the labor cost of individual recipes. The cost of the food preparation

labor is likely to have a significant impact on the menu and food system adopted

for a Mars exploration mission. Obviously, certain additional factors would have

to be accounted for, such as shifts in task durations due to the lower Mars gravity,

or the substitution of flight- qualified kitchen equipment for the standard

commercial equipment used in the project. The utility of the data acquired in this

project will depend on adjustments for differences between transit times in the test

kitchen and the Mars habitat galley. Still, ground simulation such as conducted in

this study represents a logical first step in making labor time estimates for life

23

support activities.

INSERT APPENDIX A AROUND HERE

24

ACKNOWLEDGEMENTS

Support of this work by the National Aeronautics and Space Administration under

NASA Grant NAG 5-4222 and the GE Faculty of the Future program for

undergraduate research is gratefully acknowledged.

25

BIOGRAPHICAL SKETCHES

Ammar Olabi is currently a post-doctoral fellow in the Food Science Department

at Cornell University. He completed his B.S. in Nutrition and Dietetics and M.S.

in Food Science at the American University of Beirut. In August 1995 he enrolled

in the Food Science and Technology Ph.D. program at Cornell University. He

conducted his Ph.D. research, focused on the optimization of a bioregenerative

diet, under the supervision of Dr Jean Hunter.

Jean Hunter is an associate professor in Cornell’s Department of Biological and

Environmental Engineering. She holds a B. S. from MIT and M.S. and D. Eng.

Sci. degrees from Columbia University, all in chemical engineering. Her research

interests include food engineering, fermentation and enzyme technology,

downstream processing, and the production of value-added products from food

and agricultural residues. She has taught bioprocessing, bioseparations, a

biological engineering laboratory course, and with Dr. Jackson, the undergraduate

group research project on measurement of food preparation labor which

developed the data for this paper.

26

Peter Jackson is an associate professor in the School of Operations Research and

Industrial Engineering, Cornell University. He received a B.A. in Economics from

the University of Western Ontario (1975), a M.Sc. in Statistics (1978) and a Ph.D.

in Operations Research (1980) from Stanford University. He conducts research in

production planning, resource scheduling, and inventory management and teaches

courses in industrial systems analysis, design, and engineering. He is an innovator

in experiential learning and the recipient of several teaching awards.

Michele Segal earned her B.S. at Florida State University in Psychology with a

specialty in Environmental Psychology. She completed a M.S. in Human Factors

at Cornell University, with a thesis on the use of active noise reduction headsets

to improve speech intelligibility in the space shuttle mid-deck noise environment.

She continued her work on noise and habitability after joining Johnson

Engineering Corp., Houston, TX, in the International Space Station Acoustics

group at NASA/ Johnson Space Center. Ms. Segal currently works in the Lighting

Environment Test Facility at NASA/JSC and handles lighting and human

performance issues aboard the space station.

27

Rupert Spies is a senior lecturer at the School of Hotel Administration, Cornell

University, where he teaches courses in restaurant food production, restaurant

management and catering. Besides his interest in food systems for NASA, he has

also done research on multi unit restaurant management in Germany, and

provided consulting in the areas of purchasing management, menu development

and catering, and co-taught with Dr. Hunter and Dr. Jackson, the undergraduate

group research project on measurement of food preparation labor. Prior to

receiving his degree from the Technische Universitaet Berlin, he apprenticed and

worked as a chef in various restaurants in Germany and abroad.

28

REFERENCES

1. Aft, L.S. Productivity Measurement and Improvement. Reston: RestonPublishing Company; 1983, 429 p.

2. Barnes, R.M. Motion and Time Study Design and Measurement of Work.New York: John Wiley and Sons, 7th ed.; 1980, 689 p.

3. Blanken, H.M. A time-task allocation method for labor control in hospitalfood service departments. In: Productivity Improvements in HospitalDietary Systems: an Examination of Case Studies. Proceedings of a ForumAnaheim, CA September 14-15. Chicago: Center for HospitalManagement Engineering, American Hospital Association; 1978, 222 p.

4. Bonini, K.I. Application of Work Sampling Technique to Evaluate theStaffing Pattern in a Specific Production Unit of a Hospital DietaryDepartment. Masters thesis. Columbus: Ohio State University; 1966, 62 p.

5. Donaldson, B. Productivity of dietary personnel: work samplingmethodology manual / institution management personnel, InstitutionManagement Laboratory. Madison: University of Wisconsin, College ofAgriculture, Department of Foods and Nutrition, Institution ManagementLaboratory; 1967, 106 p.

6. Drysdale, A., Hanford, T. Systems Modeling and Analysis Program,Baseline Values and Assumptions Document, CTSD-ADV-371, JSC39317. NASA-Johnson Space Center; June 99.

7. Failing, R.G., Janzen, J.L., and L.D. Blevins. Improving ProductivityThrough Work Measurement: A Cooperative Approach. New York:American Institute of Certified Public Accountants; 1988, 31 p.

8. Frabble, F. Jr. Enhance operational efficiencies with improvedergonomics. Nation’s Restaurant News August 19:92, 1996.

9. Lane, H.W. Presentation to Food Technology Commercial Space CenterBidder’s Meeting. Houston, TX; June 12 1998.

29

10. Levri, J.A., Vaccari, D.A., and A.E. Drysdale. Theory and Application ofthe Equivalent System Mass Metric. International Conference OnEnvironmental Systems, July 2000, Toulouse, France. SAE Paper 2000-01-2395.

11. Hoffman S.J., Kaplan, D.L., eds. Human Exploration of Mars: TheReference Mission of the NASA Mars Exploration Study Team; July1997. http://www-sn.jsc.nasa.gov/marsref/default.html.

12. Hunter, J., Olabi, A., Spies, R., Rovers A., Levitsky, D. Diet design andfood processing for bioregenerative life support systems. SAE 1998Transactions 107, Journal of Aerospace, Section I: 296-307, 1999.

13. Karger, D.W., Bayha, F.H. Engineered Work Measurement. New York:The Industrial Press, 2nd ed.; 1966, 722 p.

14. Krajewski, L.J., Ritzman, L.P. Operations Management: Strategy andAnalysis. Reading: Addison-Wesley, 3rd ed.; 1987, 783 p.

15. Meadows, E. How three companies increased their productivity. Fortune,Time Inc March 10: 92, 1980.

16. Montag, G.M. The feasibility of using predetermined motion times as atool in the food production management. Thesis, Ames; 1963, 110 p.

17. Niebel, B.W. Motion and Time Study. Homewood: Richard D. Irwin Inc;1976, 719 p.

18. PR Newswire Association Inc. Management science helps Disney profitsby keeping guests satisfied and happy. PR Newswire April 27, 1992.

19. Ripley, H.L. A Review of Work Simplification in Institution Management,and its Application to Three Areas of Work in a College Cafeteria.Masters thesis, Ithaca: Cornell University; 1947, 154 p.

20. Sanford, J.P. An Application of Work Sampling Technique to theMeasurement of the Job Activity of Three Food Service Managers.Masters thesis, Ithaca: Cornell University; 1962, 70 p.

21. Segal, M. May 1998 JSC COOP/Internship Final Report.

22. Shell, R.L., ed. Work Measurement Principles and Practices. Atlanta:Industrial Engineering and Management Press; 1986, 320 p.

30

23. Simounet, S.E. The Application of Scientific Management Principles tothe Study of Two Specific Problems Concerning Food Service at CarlotaMatienzo Residence Hall Dining Room, University of Puerto Rico.Masters thesis, Ithaca: Cornell University; 1957, 112 p.

24. Stenger, M.A. Team Approach to Food Service in a Decentralized DietaryDepartment. Masters thesis, St. Louis: Saint Louis University; 1964, 51 p.

25. Swift, J.A. Introduction to Modern Statistical Quality Control andManagement. Delray Beach: St. Lucie Press; 1995, pp 115-121.

26. Whitmore, Dennis A. Work Measurement. London: William HeinemannLtd.; 2nd ed., 1987, 441 p.

27. Woolley, J.H. Productivity Relationships of Hospital DietaryDepartments. Masters thesis, Madison: University of Wisconsin; 1964, 68p.

31

TABLE 1. Examples of typical tasksTask Action Definition Start of Task End of TaskAdd To put an ingredient into a vessel.

Does not include measuring.Picking up utensil orcontacting ingredient.

Releasing utensil oringredient.

Bag To put ingredients in a bag. Getting the ingredient orpicking up the bag,whichever occurs first.

Putting bag down.

Bake To cook food item in the oven orin the bread machine.

Opening oven door to putitem in / pressing startbutton on bread machine.

Closing door after removingitem from oven or removingbread from bread machine.

Blend To prepare a food item in the foodprocessor. Includes chopping,pureeing, mixing, grinding, andkneading of dough in foodprocessor. (includes checking tosee if blending is done)

Securing food processor lidon central unit.

Pressing off button if thefood item is not immediatelyremoved.

32

TABLE 2. Typical task listSoy Wheat Crepes Starts on November 12 at 3:40:00 PMTask Object Time started Time Ended Time Elapsed CommentsRead recipe 3:55:11 3:55:19 0:00:08 p.m.Get strawberries 3:55:19 3:55:20 0:00:01Transit 3:55:20 3:55:21 0:00:01Put strawberries 3:55:21 3:55:22 0:00:01Chop strawberries 3:55:22 3:59:30 0:04:08Transit 3:59:30 3:59:31 0:00:01Wash bowl 3:59:31 3:59:32 0:00:01 Get bowlTransit 3:59:32 3:59:34 0:00:02Wash bowl 3:59:34 3:59:38 0:00:04Transit 3:59:38 3:59:40 0:00:02Wash bowl 3:59:40 3:59:42 0:00:02 Put bowl

33

TABLE 3. Ten longest tasks for Seitan Stir Fry recipe (sum of all occurrences of the task within the recipe)

TASK RANK DURATION (seconds)Dead time 1 360.5Cut pepper 2 326.5Transit 3 261.5Stir 4 172.5Cut onion 5 130.5Stir vegetables 6 116.5Core pepper 7 99.5Chop seitan 8 89.5Read recipe 9 83.5Bag peppers 10 74.5

34

APPENDIX A

List Of Task Actions

Task Action

AddBagBakeBlendBoilCheckChopCleanCleanup (several cleaningtasks in sequence)CloseCookCoreCoverCrushCubeCutDead TimeDiscardDivideDrainEmptyFillFlipGetGreaseKneadLoad

Task Action

Make (Pasta, i.e. mix dough)MeasureOpenPeelPreheatPressPutReadRearrangeRiseRollSet upShakeSoakSortSpreadStirTasteTearTransitTrimTurn onTurn offWashWeighWipeWriteMiscellaneous Tasks

35

FIGURE 1. Database structure, showing tables, variables within tables andrelationships among the tables.

FIGURE 2. Major components of recipe tasks: active time, passive time, deadtime, transit, weighing and writing for 5 recipes: Ginger Melon, Seitan Stir Fry,Whole Wheat Pasta, Peanut Wheat Hot Cereal, Soy Wheat Crepes, Asian Pesto.

FIGURE 3. Major components of recipe tasks: active time, passive time, deadtime, transit, weighing and writing for 10 recipes: 1= Potato Onion Bread, 2 = DillPotatoes, 3 = Tofu Cottage Cheese, 4 = Scrambled Tofu, 5 = Cinnamon PeanutRolls, 6 = Vegetable Pasta Salad, 7 = Jerk Tofu, 8 = Chocolate Sauce withStrawberries, 9 = Broiled Zucchini with Herbs, and 10 = Cold Beet Borscht.

FIGURE 4. Average durations (seconds) across all recipes for quantityindependent tasks. 1 = Put, 2 = Discard, 3 = Turn Off, 4 = Get, 5 = Taste, 6 =Close, 7 = Rearrange, 8 = Open, 9 = Drain, 10 = Read, 11 = Clean, 12 = Check,13 = Set Up, 14 = Write, 15 = Wipe, 16 = Turn On, 17 = Miscellaneous, 18 =Empty, 19 = Cleanup, 20 = Sort.

FIGURE 5. Average durations (seconds) across all recipes for quantity dependenttasks. 1 = Transit, 2 = Cook, 3 = Cover, 4 = Shake, 5 = Weigh, 6 = Bag, 7 = Fill,8 = Grease, 9 = Rise, 10 = Add, 11 = Measure, 12 = Wash, 13 = Soak, 14 =Blend, 15 = Spread, 16 = Stir, 17 = Load, 18 = Knead, 19 = Cut, 20 = Tear, 21 =Roll, 22 = Flip, 23 = Core, 24 = Press, 25 = Crush, 26 = Peel, 27 = Chop, 28 =Divide, 29 = Cube, 30 = Preheat, 31 = Make, 32 = Boil, 33 = Trim, 34 = Bake.

FIGURE 6. Cumulative distribution function for 3 common quantity independenttasks: drain, check and close.

FIGURE 7. Pareto analysis of the task average durations (seconds) for Seitan StirFry. Each bar corresponds to a task-object combination.

FIGURE 8. GANTT chart for the production of 5 recipes. Kale Soup (dark gray),Potato Salad (light gray), Oven Roasted Vegetable Sandwich (black), Jerk Tofu(white), and Strawberry-Pineapple Drink (darkest gray). The arcs indicate the tasksequences. The workstations are listed on the y-axis.

FIGURE 9. Time use (as a percentage) of the kitchen workstations (stove, oven,counter, sink, fridge, freezer, food processor, and blender) for the production of 5Recipes: Kale Soup, Potato Salad, Oven Roasted Vegetable Sandwich, Jerk Tofu,and Strawberry-Pineapple Drink.

34

35

0

10

20

30

40

50

60

70

80

90

GingerMelon

Seitan StirFry

WholeWheatPasta

PeanutWheat Hot

Cereal

Soy WheatCrepes

AsianPesto

Recipe

Dur

atio

n (m

inut

es)

WriteWeighTransitDead TimePassive TimeActive Time

36

0

10

20

30

40

50

60

70

80

90

100

110

120

1 2 3 4 5 6 7 8 9 10

Recipe

Dur

atio

n (m

inut

es)

WRITEWEIGHTRANSITDEAD TIMEPASSIVE TIMEACTIVE TIME

37

0

10

20

30

40

50

60

70

80

90

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Task

Mea

n D

urat

ion

(sec

onds

)

38

0

100

200

300

400

500

600

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

Task

Mea

n D

urat

ion

(sec

onds

)

39

0

10

20

30

40

50

60

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58

Number of Observations

Dur

atio

n (S

econ

ds)

CLOSE CHECK DRAIN

40

0

50

100

150

200

250

300

350

400

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81

Task Number

Ave

rage

Tas

k Ti

me

(sec

onds

)

41

Stove Oven Counter Sink Fridge Freezer Food Pr. Blender

42

0%

10%

20%

30%

40%

50%

60%

70%

Stov

e

Ove

n

Cou

nter

Sink

Frid

ge

Free

zer

Food

Proc

esso

r

Blen

der

Kitchen Workstations

Perc

enta

ge U

se